Taking forecasts based on macro variables as the benchmark，this paper investigates the forecast ability of technical indicators based on moving average，momentum and on-balanced volume average strategy to directly forecast commodity future prices in China． Empirical results show that technical indicators do exhibit significant in-sample and out-of-sample forecasting power，clearly exceeding those of auto-regression model and macroeconomic variables which are widely used． Moreover，the predictive powers of technical indicators reveal robustness for different model specifications and data frequencies． Furthermore，technical indicators manifest substantial economic value for asset allocation，in terms of superior commodity risk premium forecasts and sizable utility gains． These results can provide empirical and strategic support for commodity investment and risk management．